Attribute Reduction Algorithm Of Neighborhood Rough Sets And Its Application In Classifier

With the advent of the era of big data,companies are paying more and more attention to the value of data,collecting data,establishing their own data warehouses,and obtaining effective decision support through the processing of data to bring considerable benefits to enterprises.However,as the amount of data continues to increase,the cost of data processing continues to increase,and as data processing is increasingly focused on timeliness,too long a processing process will reduce the value of the data.Therefore,how to quickly and efficiently acquire knowledge in massive data has become a hot topic in current research.As an extension to the traditional rough set,the neighborhood rough set compensates for the drawbacks that the traditional rough set can only deal with discrete data.It directly processes continuous data and reduces the data distortion caused by data discretization.However,this also brings about the problem of large calculations and long processing time.Therefore,in order to make the information processing more efficient,this paper will use the theoretical knowledge of rough sets as the basis to improve the reduction of the time overhead and apply it to the classifier.Therefore,this paper mainly made the following points(1)Analyze the existing attribute reduction algorithm,aiming at the lack of calculation volume,propose an improved voting attribute importance definition.Then,a new fast attribute reduction algorithm based on voting attribute importance is proposed.The algorithm redefines the solution to the importance of attributes and reduces the time to solve the importance of attributes.After experimental verification of multiple data sets,the algorithm is effective in reducing computation time and improving computational efficiency.(2)Analyze the research status of the existing classification algorithms,choose ID3 algorithm as an improved algorithm,and propose a decision tree algorithm based on voting attribute importance.By setting the voting attribute importance as the selection standard of the classification node,the ?-confidence level to control the size of the decision tree.After experimental verification of multiple data sets,the algorithm has improved both in the accuracy of classification and in the control of the size of the decision tree.(3)Finally,the paper summarize the proposed algorithms.